When I mentioned to people what I was reading, they'd naturally want to know what Douglas Hofstadter's classic book was about, and honestly, I never fWhen I mentioned to people what I was reading, they'd naturally want to know what Douglas Hofstadter's classic book was about, and honestly, I never felt like I was able to properly explain what was going on, even though I following the author's central thesis well enough. Sort of. I realized that an analysis of my ability to get the book, while simultaneously not being able to satisfactorily explain it to other people would probably be a fitting anecdote to give a sense of the points Hofstadter is trying to make.
Anyway, having now finished Gödel, Escher, Bach, I think I can sort of put into words what the book actual does.
What it does:
1) The author gives a primer on computer programming. How you can begin with the 0's and 1's of machine code, and through a stacking sequence of translator programs, each utilizing the structure one level down to build something more complicated one level up. The idea is that no matter how magical something looks at its highest level (think of Siri understanding your voice commands to set an alarm or initiate a web search), if you're willing to trace the functions backward down to their most fundamental level (a task of unfathomable tedium), you'll see that everything is basically just a series of those original o's and 1's.
2) The author gives a primer on biological development. How you can begin with atoms and molecules and DNA, and through a stacking sequence of enzymes and proteins, each utilizing the structure one level down to build something more complicated one level up. The idea is that no matter how magical something looks at its highest level (think of the emotion you feel when you hear an emotional piece of music), if you're willing to trace the functions backward down to their most fundamental level (a task exponentially more tedious than going back to the 0's and 1's of a computer program), you'll see that everything is basically just a series of atoms and molecules obeying the laws of physics and chemstry.
3) The author shows that while there are some startling similarities between computer programming and human neural development, humans have this extra-special something that qualitatively sets them apart from computers, and that figuring out what that extra-special something is is likely the key to creating true Artificial Intelligence.
Because that's kind of what this book is actually about: trying to understand the human neurological system in terms of computer programming (while at the same time trying to understand computer programming in terms of human neurological systems), all with the ultimate goal of trying to understand whether Artificial Intelligence is a thing we can ever truly create.
And all of this is approached from a fascinatingly tangled set of perspectives represented by the technical compositional virtuosity of J.S. Bach, the paradoxical imagery of M.C. Escher, the mathematics-shattering insight of Kurt Gödel's Incompleteness Theorum, the possibilities and limitations of Number Theory, and an assortment of silly parables featuring Achilles and Tortoise (characters from a Lewis Carroll story). Hofstadter hits on the themes ranging from Zen Buddhism, to logical recursion and self-reference, to chess strategy, to the possible musical tastes of aliens.
Written in 1979, Gödel, Escher, Bach may contain some outdated information. The author, for example, ponders in 1979 if one day an AI program could ever beat a Chess grandmaster. Meanwhile, 35+ years later, an AI had already moved on to defeating a Go grandmaster.
However, rather than feeling dated, Hofstadter's book feels eminently modern as it delves into many issues that AI developers are likely still struggling with as they continue to move forward in the field. Hell, it deals with issues that philosophers still struggle with. That is, what is consciousness? Are humans just very complicated robots? Is self-awareness just an illusion, a story we tell ourselves since we're unable to adequately comprehend the complicated neural mechanisms that control us? Or does that very complexity lead to such an impossibly large number of potentials and outcomes, that we actually do control the neural mechanisms as much as they control us?
A very fascinating read, presented in a mostly* readable fashion.
[*'Mostly' because there are several sections of the book that dig pretty deeply into number theory and programming languages that this reader found it hard to follow sometimes. I always got the gist, but I'd be lying if I followed the line-by-line logic sometimes.]...more
Loading...
